From Anonymity to Identification A

University of Miami Law School University of Miami School of Law Institutional Repository Articles Faculty and Deans 2015 From Anonymity to Identification A. Michael Froomkin University of Miami School of Law, [email protected] Follow this and additional works at: https://repository.law.miami.edu/fac_articles Part of the Privacy Law Commons, and the Science and Technology Law Commons Recommended Citation A. Michael Froomkin, From Anonymity to Identification, 1 Journal of Self-Regulation and Regulation 120 (2015). This Article is brought to you for free and open access by the Faculty and Deans at University of Miami School of Law Institutional Repository. It has been accepted for inclusion in Articles by an authorized administrator of University of Miami School of Law Institutional Repository. For more information, please contact [email protected]. Journal of Self-Regulation and Regulation Volume 01 (2015) From Anonymity to Identification A. Michael Froomkin Abstract This article examines whether anonymity online has a future. In the early days of the In- ternet, strong cryptography, anonymous remailers, and a relative lack of surveillance cre- ated an environment conducive to anonymous communication. Today, the outlook for online anonymity is poor. Several forces combine against it: ideologies that hold that anonymity is dangerous, or that identifying evil-doers is more important than ensuring a safe mechanism for unpopular speech; the profitability of identification in commerce; government surveillance; the influence of intellectual property interests and in requiring hard- ware and other tools that enforce identification; and the law at both national and supra- national levels. As a result of these forces, online anonymity is now much more difficult than previously, and looks to become less and less possible. Nevertheless, the ability to speak truly freely remains an important 'safety valve' technology for the oppressed, for dissidents, and for whistle-blowers. The article argues that as data collection online merg- es with data collection offline, the ability to speak anonymously online will only become more valuable. Technical changes will be required if online anonymity is to remain possible. Whether these changes are possible depends on whether the public comes to appre- ciate value the option of anonymous speech while it is still possible to engineer mecha- nisms to permit it. Keywords Internet, Privacy, Anonymity, Surveillance, Speech © A. Michael Froomkin DOI: 10.11588/josar.2015.0.23480 A. Michael Froomkin From Anonymity to Identification A. Michael Froomkin 1 Introduction This is not a cheerful paper, which mirrors the lecture from which it is adapted.1 It is not cheerful because the prognosis for effective online anonymity has become pro- gressively less and less cheerful. By anonymity, I mean something strong: the ability to speak without anybody being able to identify you. This untraceable anonymity is the highest level of technical protection for speech (cf. Froomkin 1995). That it is untraceable brings with it the ability to speak without fear of jail, harm, or other retaliation. This strong version of anonymity is controversial. The arguments for why untraceable anonymity is a good thing in- clude the idea that it contributes to human flourishing; people want to experiment, and the ability to experiment with less fear contributes to human self-realization. In places that are less free, avoiding retribution for saying the wrong thing may be a mat- ter of life and death. Political dissidents, ethnic minorities, religious splinter groups, people campaigning for women’s rights or gay rights, and many others are, or have been, subject to the risk of genuine and very palpable violence. If they wish to speak or write for their causes they need a means to protect themselves. Anonymity is one such tool. For those of us who do not face the danger of personal violence in retaliation for our writings, there are nonetheless other more subtle dangers. One is the danger of profiling. Profiling enables stereotypical discrimination on the basis of sexual orienta- tion, political opinion, or other characteristics. Less serious, but annoying, is the effort commercial entities make to profile for marketing purposes. Anonymity is a way to defend against that profiling, a protection that is desirable not just because some find it creepy or uncomfortable to be profiled, but also because the people doing the profiling could actually exercise market power by doing price discrimination (DeLong et al. 2000). When the Internet started, one byproduct of the architecture of the Internet was that online anonymity was easy to achieve. It required only minimal technical knowledge, or fairly simple tools, or assistance from the right people. In those days you 1 The lecture from which this paper is adapted was delivered at the University of Heidelberg on Dec 11, 2015, under the auspices of the Netzpolitik AG of the University of Heidelberg. I am very grateful to Dr. Wolf J. Schünemann and all my hosts for the invitation and the warm welcome. Unless other- wise noted, this paper attempts to reflect legal and technical developments up to May 1, 2015. 121 From Anonymity to Identification could be anonymous and have a great deal of faith in being successfully untraceable. Cryptography made this possible. PGP – "Pretty Good Privacy" – was one of the early and important tools, perhaps the first consumer-oriented cryptography. Another very important tool was the secure anonymous remailer (cf. Froomkin 1997: 129). An anonymous remailer works as follows: Alice sends out an email destined for Bob which contains within it one or more layers. Each layer consists of an encrypted message, which I will call the payload, and unencrypted delivery instructions, making the entire message something like a sealed paper letter with an address written on the front. Although destined for Bob, Alice addresses the message to a remailer operator. The remailer operator strips off the headers of Alice's email – the part that identified it as coming from her – and forwards the payload message as directed. Alice could direct that the payload go to a second remailer, who would decrypt the payload only to find in it another layer of address and (to him) unreadable payload. By chaining the message in this manner through two or more remailers, Alice could ensure that before the message reached Bob its origins would be thoroughly obfuscated. What is more, if Al- ice and Bob used encryption to exchange messages, then none of the remailer operators would ever know what Alice was saying. Even if Bob was not so sophisticated, only the last remailer in the chain would be capable of reading the cleartext, and that per- son would have no way of finding out who originally sent the message unless every remailer in the chain kept logs and cooperated in unraveling the message's path. Sending messages this way was never easy, and would be harder today because there are fewer reliable remailers. Remailers ran into two serious problems. The first is that spammers abused the remailer network. Strangely, a small amount of spam is a good thing for the remailer network because the spammers can be relied on to create a certain volume of traffic and this makes it harder for any observer to trace the genuine messages as they move through the network (ib.). But in short order the volume of spam increased to the point that so much spam went through the network as to first burden it and then choke many of the remailers entirely (Canter 2003). Worse, the remailers faced a serious legal problem. The end point in any chain of remailers – the exit node – carries legal risk for misuse of the network. If, for example, your computer was used to send a death threat to the US President, ignorance of the message's content was not a comforting defense, and certainly was no shield against an investigation. To say that the entire process was automated did not necessarily pro- vide a sufficient defense either.2 As governments and police departments became more technologically savvy, the email operators increasingly decided it was just too unsafe to run a remailer, or at least to run an exit node. As fewer and fewer remailers were willing to be exits, the ones that persisted became overwhelmed with all the 2 Notably, the protections of the CDA against civil liability for third-party postings do not apply to criminal law. 47 U.S.C. §230(e)(1). 122 A. Michael Froomkin spam that sought release into the greater Internet, thus accelerating the network's death spiral. Almost all these anonymity systems depended on some kind of cryptography. Western governments, and in particular the U.S. government, worked very hard to slow the spread of consumer cryptography. Governments were basically able to re- strict the spread of cryptography, originally through export control (Froomkin 1996: 15–75). By preventing standardization, they made it easier to maintain the wiretapping and interception capabilities of both national security services and ordinary law en- forcement. For many years this project of prevention was successful, and consumer cryptography was rare; indeed, only specialists used it – the presence of strongly encrypted traffic was sufficiently unusual to raise the possibility that observers might use it as a signal that one had something to hide, drawing extra attention from the authori- ties. All these things undermine access to anonymity (in the strong sense). We can di- vide the history of Internet communications into three periods when using anonymity as a yardstick. In the first period, as noted above, anonymity was easy if you knew what you were doing, but it was not available easily to the consumer. This period end- ed when the remailers died off and strong anonymity was reduced to what I would call a 'safety valve' technology.

Load more